def test_make_hotshot_network_small_dataset():
nn = make_hotshot_network(
peptide_length=3,
n_amino_acids=2,
activation="relu",
init="lecun_uniform",
loss="mse",
layer_sizes=[4],
optimizer=RMSprop(lr=0.05, rho=0.9, epsilon=1e-6),
batch_normalization=False)
X_binary = np.array([
[True, False, True, False, True, False],
[True, False, True, False, False, True],
[True, False, False, True, True, False],
[True, False, False, True, False, True],
[False, True, True, False, True, False],
[False, True, True, False, False, True],
], dtype=bool)
Y = np.array([0.0, 0.0, 0.0, 0.0, 1.0, 1.0])
nn.fit(X_binary, Y, nb_epoch=20)
Y_pred = nn.predict(X_binary)
print(Y)
print(Y_pred)
for (Y_i, Y_pred_i) in zip(Y, Y_pred):
if Y_i:
assert Y_pred_i >= 0.6, "Expected higher value than %f" % Y_pred_i
else:
assert Y_pred_i <= 0.4, "Expected lower value than %f" % Y_pred_i
def test_make_hotshot_network_small_dataset():
nn = make_hotshot_network(peptide_length=3,
n_amino_acids=2,
activation="relu",
init="lecun_uniform",
loss="mse",
layer_sizes=[4],
optimizer=RMSprop(lr=0.05, rho=0.9,
epsilon=1e-6),
batch_normalization=False)
X_binary = np.array([
[True, False, True, False, True, False],
[True, False, True, False, False, True],
[True, False, False, True, True, False],
[True, False, False, True, False, True],
[False, True, True, False, True, False],
[False, True, True, False, False, True],
],
dtype=bool)
Y = np.array([0.0, 0.0, 0.0, 0.0, 1.0, 1.0])
nn.fit(X_binary, Y, nb_epoch=200)
Y_pred = nn.predict(X_binary)
print(Y)
print(Y_pred)
for (Y_i, Y_pred_i) in zip(Y, Y_pred):
if Y_i:
assert Y_pred_i >= 0.6, "Expected higher value than %f" % Y_pred_i
else:
assert Y_pred_i <= 0.4, "Expected lower value than %f" % Y_pred_i
def make_model(
config,
peptide_length=9):
"""
If we're using a learned vector embedding for amino acids
then generate a network that expects index inputs,
otherwise assume a 1-of-k binary encoding.
"""
print("===")
print(config)
if config.embedding_size:
return make_embedding_network(
peptide_length=peptide_length,
embedding_input_dim=20,
embedding_output_dim=config.embedding_size,
layer_sizes=[config.hidden_layer_size],
activation=config.activation,
init=config.init,
loss=config.loss,
dropout_probability=config.dropout_probability,
learning_rate=config.learning_rate,
optimizer=config.optimizer)
else:
return make_hotshot_network(
peptide_length=peptide_length,
layer_sizes=[config.hidden_layer_size],
activation=config.activation,
init=config.init,
loss=config.loss,
dropout_probability=config.dropout_probability,
learning_rate=config.learning_rate,
optimizer=config.optimizer)
def test_make_hotshot_network_properties():
layer_sizes = [3, 4]
nn = make_hotshot_network(peptide_length=3,
n_amino_acids=2,
activation="relu",
init="lecun_uniform",
loss=mse,
layer_sizes=layer_sizes,
batch_normalization=False,
optimizer=RMSprop(lr=0.7, rho=0.9, epsilon=1e-6))
eq_(nn.layers[0].input_dim, 6)
eq_(nn.loss, mse)
assert np.allclose(nn.optimizer.lr.eval(), 0.7)
print(nn.layers)
# since the hotshot network doesn't have an embedding layer + flatten
# we expect two fewer total layers than the embedding network.
eq_(len(nn.layers), 2 * (1 + len(layer_sizes)))
def test_make_hotshot_network_properties():
layer_sizes = [3, 4]
nn = make_hotshot_network(
peptide_length=3,
n_amino_acids=2,
activation="relu",
init="lecun_uniform",
loss=mse,
layer_sizes=layer_sizes,
batch_normalization=False,
optimizer=RMSprop(lr=0.7, rho=0.9, epsilon=1e-6))
eq_(nn.layers[0].input_dim, 6)
eq_(nn.loss, mse)
assert np.allclose(nn.optimizer.lr.eval(), 0.7)
print(nn.layers)
# since the hotshot network doesn't have an embedding layer + flatten
# we expect two fewer total layers than the embedding network.
eq_(len(nn.layers), 2 * (1 + len(layer_sizes)))
